Operationalizing FDT
This post is an attempt to better operationalize FDT (functional decision theory). It answers the…
Category: Opinion
Operationalizing FDT
This post is an attempt to better operationalize FDT (functional decision theory). It answers the…
Steering Awareness: Models Can Be Trained to Detect Activation Steering
TL;DRLLMs can be trained to detect activation steering robustly. With lightweight fine-tuning, models learn to…
All technical alignment plans are steps in the dark
One reason aligning superintelligent AI will be hard is because we don’t get to test…
A Plan ‘B’ for AI safety
TL;DR: Teaching AI the value of biology in solving future AI-relevant problems may serve as…
Anthropic vs USG. What will happen by May 1st? Long careful forecast.
On March 4th, 2026, the Pentagon did something it had never done to an American company…
Ideologies Embed Taboos Against Common Knowledge Formation: a Case Study with LLMs
LLMs are searchable holograms of the text corpus they were trained on. RLHF LLM chat…
Are AIs more likely to pursue on-episode or beyond-episode reward?
Consider an AI that terminally pursues reward. How dangerous is this? It depends on how…
Modeling a Constant-Compute Automated AI R&D Process
We’d like to know how much limits on compute scaling will constrain AI R&D. This…
Forecasting Dojo Meetup – Open discussion about our forecasting process
Hi Everyone, The next meetup of the forecasting practice group is here! This time we'll…